Research map / Video / Action recognition
Programmatic Concept Learning for Human Motion Description and Synthesis
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| represent joint trajectories with spline primitives instead of Motion Programs primitives | PMC motion representation · GolfDB | KD: 1.171% -> 0.529% | Helped |
| represent joint trajectories with spline primitives instead of Motion Programs primitives | PMC motion representation · MotiCon | Time: 57.65 -> 15.92 | Helped |
| use motion primitives rather than pose sequences as input | GRU-based motion description · MotiCon | SeqAcc: 81.94% -> 91.53% | Helped |
| add primitive and transition losses (ablation) | GRU-Prim motion description · MotiCon | mAP: 0.10% -> 38.74% | Helped |
| add primitive loss (ablation) | GRU-Prim motion description without primitive loss · MotiCon | mAP: 20.58% -> 38.74% | Helped |
| add transition loss (ablation) | GRU-Prim motion description without transition loss · MotiCon | mAP: 17.87% -> 38.74% | Helped |
| use CTC rather than seq2seq decoding | motion description · MotiCon | SeqAcc: 40.23% -> 91.53% | Helped |
| use spline primitives as the baseline model's output representation | JL2P controlled motion synthesis · MotiCon | APE: 0.2504 -> 0.2052 | Helped |
| sample the learned concept primitive distributions for action-conditioned synthesis | PMC · MotiCon | MM: 2.001 -> 1.141 | Mixed |
Closest papers by meaning
- Hierarchical Motion Understanding via Motion Programs (CVPR 2021 · similarity 0.90)
- HuMoCon: Concept Discovery for Human Motion Understanding (CVPR 2025 · similarity 0.89)
- Disentangled Concepts Speak Louder Than Words: Explainable Video Action Recognition (NeurIPS 2025 · similarity 0.87)
- Scaling Large Motion Models with Million-Level Human Motions (ICML 2025 · similarity 0.86)
- H-MoRe: Learning Human-centric Motion Representation for Action Analysis (CVPR 2025 · similarity 0.86)
- AvatarGPT: All-in-One Framework for Motion Understanding Planning Generation and Beyond (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.